Sam Altman, chief executive of OpenAI, has said artificial intelligence could go "very badly" in two specific ways, in a call to slow the technology's rapid development. Writing on X, Altman said the first danger would be losing control over AI's future, which he called "unacceptable," while the second would be an excessive concentration of power, in which an extraordinarily powerful AI is used by one person or company to impose their worldview on everyone else. His comments follow a stark warning from Jacob Coxon, a former researcher at both OpenAI and Anthropic, who resigned this week citing existential risks.

Altman said he is working with other executives to "pace" AI progress, stressing this does not mean halting development but ensuring safety and alignment techniques keep up with capabilities. He argued that competitive pressure should never justify recklessness. Coxon, who spent three years on pretraining research, said neither Anthropic nor OpenAI was "acting responsibly," warning that superhuman systems could soon hack any target, transform industries overnight and seize real-world power, and that many insiders privately fear AI could kill everyone by the end of the decade. Geoffrey Hinton, the so-called "Godfather of AI," has separately warned that superintelligent systems could lead to human extinction, saying there is no scientific consensus that such systems can be developed safely.